What Is Conversion Rate? How to Measure and Improve It

A store gets 10,000 visitors in a month and makes 200 sales. Another store gets half the traffic, 5,000 visitors, and also makes 200 sales. The second store is doing something right that the first one isn't, and the number that shows it is the conversion rate.

Conversion rate is the percentage of visitors who complete the action you want, whether that's a purchase, a signup, a form submission, or a download. It matters because traffic on its own doesn't pay the bills. A page that turns 2% of visitors into customers needs twice the traffic to match a page that converts 4%, and traffic costs money whether it comes from ads or organic search. Improving conversion rate gets more value out of the traffic you already have, instead of paying for more of it.

Most people know roughly what conversion rate means but calculate it wrong, compare it against the wrong benchmark, or focus on the wrong page when trying to improve it. A checkout page and a blog post shouldn't be judged by the same standard, and a "good" conversion rate for one industry can be a poor one for another.

This article covers what conversion rate actually is, how to calculate it correctly, what counts as a good rate depending on your industry and page type, and the specific changes that move the number once you know where it's stuck.

What is conversion rate?

Conversion rate is the percentage of an eligible audience that completes a defined desired action, such as buying, signing up, downloading, or booking a call. The eligible audience is the group that actually had the chance to take that action: visitors to a page, people who clicked an ad, or leads handed to sales.

A higher conversion rate means more of the people who showed interest took action. A lower one tells you where interest leaks out. That's the whole value of the metric. It points to the step that needs work.

Two kinds of conversion matter here:

• Macro-conversions: the action the business earns from, like a purchase or demo request.
• Micro-conversions: smaller steps toward it, like add-to-cart or a pricing page view.

Watch the vocabulary in your tools, too. According to Google Analytics Help, actions that matter to your business are now called "key events," while a "conversion" refers to an action you use to measure ad campaigns and optimize bidding. Same idea, different label, depending on which report you open.

How do you calculate conversion rate?

Divide the number of conversions by the eligible population, then multiply by 100.

Conversion rate = (conversions ÷ eligible population) × 100

Say 50 people buy out of 1,000 visitors. That's 50 ÷ 1,000 × 100, a 5% conversion rate.

The math is easy. The choices before the math are where teams go wrong. Work through them in this order:

1. Name the one action you're counting.
2. Pick the denominator that had a real chance to take it.
3. Set the time window, such as the last 28 days.
4. Decide whether to count unique people or every conversion event.
5. Divide conversions by the denominator and multiply by 100.

Platforms make these choices for you, so check their definitions. Google Ads Help defines conversion rate as "the average number of conversions per ad interaction," calculated by dividing conversions by total eligible interactions. That's a click-based denominator, not a visitor-based one. It's also the reason ad platform rates can pass 100%, which gets its own section below.

What should count as a conversion?

Count the action that proves progress toward the goal of that specific page or campaign, and nothing softer. A webinar page converts when someone registers, not when they scroll to the bottom.

Common conversion types include:

• Purchase or completed checkout
• Account creation or free trial start
• Email or newsletter sign-up
• Content download, such as a report or template
• Form fill, such as a demo or quote request
• App install
• Booked call or meeting

The clearest pages make the conversion obvious. Shopify's free trial page asks for an email address and little else, so the conversion is plainly "trial started." Asana's account creation page does the same with a single work email field, which also filters toward business users. Marketing Brew places its subscription field right next to its value proposition for marketers, so a sign-up is the one thing the page asks for.

Notice what those pages have in common. Each page presents one clear primary action. One page, one conversion, one number worth tracking.

Which denominator should you use?

Use the group that had a real chance to convert, for the exact question you're answering. Most arguments about conversion rate are really arguments about the denominator.

Your question	Numerator	Denominator	Watch out for
Is this ad sending people who act?	Conversions	Ad clicks or interactions	Platform counting rules (one vs. every)
Is this landing page persuading?	Unique converters	Unique visitors to that page	Bots, internal traffic, repeat visits
Is checkout working?	Orders	Sessions that reached checkout	Mixing in sessions that never saw checkout
Is this email doing its job?	Clicks or conversions	Delivered emails	Using "sent," which includes bounces
Are our leads any good?	Qualified leads	All new leads	Changing qualification rules mid-period
Is sales closing?	Closed-won deals	Opportunities created	Comparing deals to leads from a different month
Is the trial converting?	Paid accounts	Trial starts in the same cohort	Counting upgrades from older trials
Is the campaign paying back?	Customers or revenue	Ad spend	Stopping at the lead stage

A cohort is a group of people who entered at the same time, like everyone who started a trial in March. Cohorts keep the numerator and denominator talking about the same people.

Can a conversion rate exceed 100%?

Yes, and most introductory guides omit the reason: the difference between counting conversion events and counting people. A rate passes 100% when the numerator counts events, and the denominator counts something smaller, such as clicks or people. It can't pass 100% when both sides count the same unique people. If it does, something in your tracking is broken.

a. Conversion events vs. unique converters

A conversion event is each time the action fires. A unique converter is a person who fired it at least once. One person who downloads three ebooks is one converter and three events.

Because Google Ads Help defines conversion rate as the average number of conversions per interaction, the average climbs above one as soon as a single click can lead to several conversions; say 100 clicks bring in 60 buyers who place 150 orders between them. Count every order, and the rate is 150%. Count one conversion per click, and it's 60%.

Neither is wrong. They answer different questions:

• Lead-gen page: count unique people.
• Ecommerce revenue: count every order.
• Ad bidding: match whatever the bid strategy optimizes for.

b. Repeat conversions

Returning customers who buy again add conversions without adding new people. Over a long window, a loyal customer base can lift an event-based rate well above what first-time visitors do. Split new and returning visitors so a weak acquisition page can't hide behind repeat buyers.

c. Denominator choice

The same numerator gives different rates depending on what you divide by. Say your store has 1,000 users who make 1,600 sessions and place 80 orders in a month. Per user, that's 8%. Per session, it's 5%.

Pick one, write it on the report, and keep it. A rate that switches from users to sessions between months will show a drop that never happened.

When over 100% means something broke

If you count unique converters against unique visitors and still land above 100%, check for:

• A tag firing twice on one thank-you page
• CRM conversions imported with no matching visit
• A numerator window longer than the denominator window

One campaign, five different conversion rates

Here's a hypothetical B2B campaign to show why a single campaign has several valid conversion rates. Say you spend $10,000 on search ads for a demo offer over 30 days.

Stage	Count	Rate	What it measures
Ad clicks	2,000	n/a	Traffic bought
Unique landing page visitors	1,500	75% of clicks	Clicks that became real visits
Demo form fills (unique)	120	8% of visitors	Landing page conversion rate
Qualified leads	48	40% of form fills	Lead quality
Opportunities	12	25% of qualified leads	Sales acceptance
Closed-won deals	3	25% of opportunities	Sales conversion rate

From the same hypothetical campaign, you can also report 6% (form fills per click, what an ad platform would show) and 0.15% (deals per click). Every one of these numbers is correct.

The trap is comparing them. An 8% landing page rate is not "worse" than a 25% close rate. They measure different people at different stages. Put 8% next to 25% in a slide, and someone will ask why marketing underperforms sales. The honest comparison is each stage against its own history.

i. Landing page conversion rate

This is conversions divided by unique visitors to one page. It tells you whether the page persuades the people who arrive. It says nothing about whether those people buy later. For help deciding whether a campaign needs its own page at all, see campaign landing page vs. website page.

ii. Sales conversion rate

This is closed-won deals divided by leads or opportunities in the same cohort. Pick one starting point and keep it. Lead-to-close and opportunity-to-close answer different questions, and switching between them makes trends meaningless.

iii. SaaS free-to-paid conversion rate

This is paid accounts divided by trial or freemium sign-ups from the same period. Say 400 people start a 14-day trial in March and 60 of them pay: that's 15%. Count only upgrades from that cohort, and give it enough time to close after the trial ends.

Conversion rate vs. click-through rate

Click-through rate (CTR) is clicks divided by impressions. Conversion rate starts after the click and measures what people do once they arrive. CTR tells you whether the ad earned attention. Conversion rate tells you whether the page and offer earned action.

A click only counts as a conversion when the click itself is the campaign goal. For most campaigns, it isn't.

Read the two together. A high CTR with a low conversion rate often means the ad promised something the page doesn't deliver. Picture a headline that says "free pricing calculator" and a page that opens on a demo form. Fix the promise, not the button color.

Why conversion rate matters for ad spend and revenue

Conversion rate sets how much you pay for every result. Cost per acquisition (CPA) is spend divided by conversions, so a better rate lowers CPA without touching the budget.

Use the hypothetical campaign above. At $10,000 and 120 form fills, you pay about $83 per lead. Lift the landing page from 8% to 10% on the same 1,500 visitors, and you get 150 leads at about $67 each. Same spend, same traffic, 30 more leads.

Revenue follows the same chain: visitors × conversion rate × average order value. Double any one of the three and revenue doubles. Buying more traffic costs money every month. A page change is paid for once.

That's why conversion rate earns a spot as a KPI. It's also why you can't track it alone. Check lead quality, revenue, order value, refunds, and retention alongside it. A gated "free gift" can push sign-ups up while qualified pipeline falls.

How conversion rate changes by channel, device, and audience

The same page converts differently depending on who arrives and how. Compare each channel against its own baseline, never against each other as one pooled number.

a. Paid search and paid social

Search visitors typed a need. Social visitors were interrupted mid-scroll and need more convincing. Tighter audience targeting sends fewer uninterested people to the page, which lifts the rate.

b. Organic search

Organic traffic mixes researchers and buyers. A blog post and a pricing page can both rank, and they should not share one conversion goal. Measure organic landing pages by the action that fits their intent.

c. Email

Use delivered emails as the denominator, not sent. Don't treat opens as a conversion signal. Clicks and on-site conversions show what people actually did.

d. Mobile vs. desktop

Mobile and desktop visitors often behave differently on the same page. Long forms, small tap targets, and slow loads hurt mobile first. Split every conversion rate report by device before you decide the page is the problem. Say desktop converts at 9% and mobile at 3%: the blended average describes nobody.

What is a good conversion rate?

A good conversion rate is one that beats your own baseline for the same stage, channel, and conversion definition. Benchmarks give you context. Your own data tells you what to change.

Here's why published benchmarks mislead so easily. Triple Whale's ecommerce benchmark roundup lists visitor-to-order rates such as 1.40% for Shopify stores, 1.71% for Sports & Outdoors, 1.81% for Apparel & Accessories, 2.56% for Health & Beauty, and 2.74% for Food & Beverage. Meanwhile, First Page Sage's sales funnel benchmark report, built from internal and anonymized client data gathered from 2017 through 2025, lists ecommerce stage rates between 23% and 66%.

Both can be right. One measures visitors who bought. The other measures movement from one funnel stage to the next. Visitor-to-purchase, visitor-to-lead, lead-to-customer, trial-to-paid, and click-to-conversion rates are not comparable, however similar the label looks.

Before you trust any benchmark, check four things:

• Which funnel stage it measures
• Which traffic mix and devices it includes
• Which period the data covers
• How it defines a conversion

If you can't answer those, treat the figure as trivia.

The practical rule: use benchmarks to sanity-check whether you're wildly off, then set targets from your own baseline, segments, and funnel evidence. Say a page moves from 3% to 4% against its own history. That's a one-third improvement, whatever an industry average says.

When is conversion-rate data misleading?

The formula can be correct while the reported rate is still misleading, because the numerator and denominator describe different populations or periods. Check these six causes before you change a page.

a. Attribution windows

An attribution window is the period after a click or view in which a conversion gets credited to it. A longer window can credit conversions that happen well after the click, while a shorter one leaves those later conversions out. So the same campaign reports a different conversion rate depending on the window. Your ad platform and your analytics tool can each apply their own window, so they'll disagree. Keep the window fixed when you compare periods, and note it on every report.

b. Cross-device behavior

Someone clicks your ad on a phone at lunch and fills the form on a laptop that evening. Without signed-in users or cross-device modeling, that's one visitor who didn't convert and another who converted from nowhere. Mobile rates look worse than they are as a result.

c. Consent loss

Where visitors can decline analytics cookies, the ones who decline can still convert. Your analytics tool just never records their visit. Both numerator and denominator shrink, and not evenly. Compare analytics conversions against form submissions in your CRM to size the gap.

d. Bot filtering

Bot filtering only removes the bots your tool recognizes. Spam form fills inflate conversions, and crawler visits inflate the denominator. Filter your own team's traffic too, especially on low-volume pages where ten internal visits move the rate.

e. Offline conversions

Deals that close by phone or in a sales meeting never touch your website tag. Import closed-won data from your CRM back into your ad and analytics platforms. Otherwise, your reports stop at the form fill.

e. Mismatched comparison periods

Compare like with like: the same weekdays, the same season, the same definition. A week with a holiday sale against a week without one tells you about the sale, not the page. The same goes for a trial cohort measured before its members have had time to pay.

Also check your data retention. The GA4 migration reference notes that retention can be set to 2 or 14 months. At 2 months, year-over-year comparisons in exploration reports aren't possible, so set 14 months before you need them.

Before you report a change in conversion rate, confirm:

1. The attribution window matches the earlier period.
2. The counting rule (every vs. one) hasn't changed.
3. Analytics conversions roughly match CRM submissions.
4. Bot and internal traffic are filtered on both periods.
5. Both periods cover the same weekdays and season.

How do you improve conversion rate?

Find the step with the largest useful drop-off, then fix the smallest point of friction inside it. That work is called conversion rate optimization (CRO): the practice of increasing the share of visitors who complete a desired action by removing obstacles and testing changes.

Start with diagnosis, not ideas:

• Set a baseline for each funnel stage over a full, comparable period.
• Segment by traffic source, device, audience, location, and landing page.
• Find the stage where the most qualified people drop out.
• Watch what happens there: recordings, form analytics, and support questions.
• Pick the smallest change that addresses the likely cause.
• Test it, or ship it and measure it against the baseline.

In the hypothetical campaign above, 500 of 2,000 clicks never became a real visit. That's a 25% loss before the page even had a chance. Page speed or tracking breaks would be the first things to check, ahead of any copy rewrite.

Once you know where the problem is, these are the fixes to try.

a. Match the message to the click

The headline should repeat what the ad or email promised, in the same words. A clear value proposition that states who it's for and what they get beats a clever one.

b. Make the call to action obvious

Use one primary call to action per page, and say what happens next: "Get my pricing" beats "Submit." Remove competing links and navigation that pull people away from the one action you're measuring.

c. Add proof where doubt appears

Put social proof and trust signals right next to the decision. Customer logos, a short testimonial, or security and privacy notes belong beside the form, not in a footer nobody reaches.

d. Cut friction from forms and checkout

Every field is a reason to leave. Ask only for what sales needs at this stage. Calendly's sign-up page shows the idea well: single sign-on options and one clear next action, so starting an account takes a couple of clicks instead of a form.

Fix mobile and speed

Check the page on a real phone. Look for forms that push the button below the fold, pop-ups that cover content, and images that load slowly on cellular connections.

If you need a plan for the whole campaign rather than one page, the guide to running a successful marketing campaign covers goals, funnel forecasts, and attribution.

When should you use A/B testing?

Use A/B testing when you have enough traffic to detect the change you care about, and you've chosen one primary metric before you start. Without both, you'll get a winner that isn't real.

An A/B test splits traffic between two versions of a page and compares their conversion rates. Common candidates are headlines, offers, calls to action, forms, layouts, and checkout steps.

i. Decide the sample size first

Evan Miller's How Not To Run an A/B Test makes the key point: if you run a test "until we see a significant difference" instead of fixing the sample in advance, "all the reported significance levels become meaningless." His rule of thumb for the sample needed per variant is roughly 16 × σ² ÷ δ², where δ is the smallest change you want to detect.

Here's an illustrative calculation. Say your page converts at 4% and you want to detect a lift to 4.8%, a 20% relative improvement. The variance is 0.04 × 0.96, or 0.0384. The difference is 0.008. That works out to 16 × 0.0384 ÷ 0.000064, or about 9,600 visitors per variant. At 500 visitors a week, that test takes about nine months. Most small pages can't support it.

That smallest change you want to detect is the minimum detectable effect. Smaller effects need far more traffic.

ii. Run full weeks and stop on schedule

Weekday and weekend visitors behave differently. Run tests in whole weeks, and avoid peak sales periods unless that's what you're testing. Don't stop early because one version looks ahead on day three.

iii. Test one element or a bigger change?

On high-traffic pages, test single elements, so you learn what worked. On low-traffic pages, test meaningful changes, such as a new offer or a shorter form, because a small tweak will never reach significance. If traffic is too thin for either, ship the change and compare against your baseline over matching periods.

Tools and approaches for measuring and improving conversion rate

Measuring conversion rate and improving it usually take more than one tool. Here's how the main approaches compare.

Approach	Good for	Watch out for
Web analytics platform	Visitor-based rates, segments, funnels	Consent loss, bot noise, retention limits
Ad platform reporting	Click-based rates, bidding	Its own attribution window and counting rules
CRM	Lead-to-deal and sales conversion rates	Only as good as the form and source data feeding it
Site CMS pages	Evergreen pages, SEO	Slow to build campaign variants without dev help
Dedicated landing page platform	Fast campaign pages, variants, built-in tests	Must connect cleanly to your CRM and analytics
Custom-coded pages	Full control	Every change waits in the dev queue

If you're starting out, pick based on how often you change pages. A few evergreen pages a year fit your CMS. Several campaigns a month, each needing its own message-matched page, is where a dedicated platform earns its place. For a deeper look at that trade-off, read whether an AI website builder can replace Webflow or Unbounce.

Whichever you choose, the page is not the outcome. It's one intervention in a loop: find the drop-off, make the smallest useful change, publish it safely, and measure what happened to conversions and qualified pipeline. Episode runs that loop in one place. It flags campaigns and pages worth improving, recommends the change, publishes on-brand pages without a dev ticket, and measures conversions alongside the lead data your CRM sends back.

For your next step, take one live campaign and fill in the denominator table above for each stage. Then run the five-point check from the misleading-data section on those numbers. The stage with the biggest drop among qualified visitors is where your next change goes.

How Marketing Teams Use Episode to Improve Campaign Conversion Rates

Episode helps marketing teams turn more of their paid clicks into leads by putting everything they need to improve a landing page in one place. Teams can match each page to the ad that sent the visitor, test new headlines and offers without touching the live page, and see in plain language where visitors drop off, down to the form field or button that loses them. They can also show different audiences the version of the page that fits them, and send leads straight to the tools they already use so follow-up stays fast. Because finding a problem, fixing it, and measuring the result all happen in one workspace, teams run more tests and learn faster what their audience responds to. Episode also includes search tools, so what works in paid campaigns can feed content that keeps bringing in visitors after the ad budget stops.